Supported Document Types¶

ChronoScope intelligently processes multiple document formats to extract timeline events.

Overview¶

ChronoScope uses AI-powered document classification to automatically detect document types and optimize extraction strategies. Understanding which document types work best helps you prepare files for optimal results.

Supported File Formats¶

Fully Supported Formats¶

Limited Support¶

Not Supported¶

PDF Processing (Dual-Library Approach)¶

How It Works¶

ChronoScope uses two PDF extraction libraries with automatic fallback:

1. PyMuPDF (primary) - Fast, reliable, works for most PDFs
2. pdfplumber (fallback) - Better for complex layouts and tables

Automatic selection:

User uploads PDF
    ↓
Try PyMuPDF extraction
    ↓
Success? → Return extracted text
    ↓
Failed? → Try pdfplumber
    ↓
Success? → Return extracted text
    ↓
Failed? → Show error (likely image-based PDF)

PDF Requirements¶

For best results, PDFs must have:

✅ Selectable text - You can highlight and copy text ✅ Structured layout - Clear sections and paragraphs ✅ Readable fonts - Standard fonts, not decorative ✅ No encryption - Unprotected files

Testing if your PDF will work:

1. Open PDF in viewer (Preview, Adobe Reader, etc.)
2. Try selecting text with cursor
3. If you can copy-paste text → ✅ Will work!
4. If text is not selectable → ❌ Need OCR first

Image-Based PDFs¶

If your PDF is a scanned image:

Option 1: Use OCR software

# Example with ocrmypdf (Mac/Linux)
brew install ocrmypdf
ocrmypdf input.pdf output.pdf

# Example with Adobe Acrobat (Windows/Mac)
File → Export To → Text (OCR if needed)

Option 2: Online OCR services

• Adobe Online OCR
• SmallPDF OCR
• PDFCandy OCR

Document Type Detection¶

ChronoScope automatically classifies documents into types for optimized extraction.

Detected Types¶

1. Resume/CV¶

Characteristics: - Section headers: "Experience", "Education", "Skills" - Date ranges (Month Year - Month Year) - Job titles and company names - Bullet points with achievements

Extraction strategy: - Focus on work experience and education sections - Extract positions, companies, dates, locations - Parse education degrees and institutions - High confidence for explicit dates

Example structure:

PROFESSIONAL EXPERIENCE

Senior Software Engineer
TechCorp Inc., San Francisco, CA
June 2020 - Present
• Led development team of 5 engineers
• Built ML pipeline processing 1M+ events/day

2. Cover Letter¶

Characteristics: - Narrative format (paragraphs) - Mentions of specific experiences - Date references within text - Personal storytelling

Extraction strategy: - Parse narrative for timeline references - Extract mentioned companies/roles - Infer dates from contextual clues - Moderate confidence (dates often implicit)

Example structure:

During my three years at Google (2018-2021), I developed
a passion for machine learning. This led me to pursue
graduate studies at Stanford in 2021...

3. Personal Statement¶

Characteristics: - Academic focus - Research mentions - Publication references - Conference presentations

Extraction strategy: - Extract research projects and publications - Parse academic institutions and degrees - Identify grants, awards, conferences - High confidence for publications (explicit dates)

Example structure:

My research on neural networks, published in NeurIPS 2022,
built upon work I conducted during my PhD at MIT (2018-2022).
I presented preliminary findings at ICML 2021...

4. General Document¶

Characteristics: - Unstructured text - Mixed content types - No clear timeline format - Informal writing

Extraction strategy: - Broad pattern matching for dates - Extract any temporal references - Lower confidence scores - Manual review recommended

Extraction Quality by Document Type¶

Resume/CV: ⭐⭐⭐⭐⭐ Excellent¶

Success rate: 90-95%

Why it works well: - Explicit date formats - Structured sections - Clear temporal order - Standardized formatting

Typical extraction: - 10-20 events per page - 85-95% confidence scores - Complete date ranges - Locations and people

Optimization tips: - Use standard section headers - Include month/year for dates - Separate entries with spacing - Add locations for each role

Cover Letter: ⭐⭐⭐⭐ Very Good¶

Success rate: 70-85%

Why it works: - Narrative timeline structure - Specific experience mentions - Clear transitions between roles

Challenges: - Dates often implicit ("three years ago") - Locations may be missing - Less structured format

Typical extraction: - 5-10 events per document - 70-85% confidence scores - Some date ranges missing - Fewer location details

Optimization tips: - Mention explicit dates when possible - Reference specific companies/institutions - Use temporal markers ("In 2020...") - Include locations in narrative

Personal Statement: ⭐⭐⭐⭐ Very Good¶

Success rate: 75-90%

Why it works: - Academic rigor with dates - Structured achievements - Publication references

Challenges: - Mixed chronological order - Research spans overlap - Complex temporal relationships

Typical extraction: - 8-15 events per document - 75-90% confidence scores - Good date coverage - Strong institutional references

Optimization tips: - Include dates for all milestones - Reference publications with years - Mention conference dates - List institutions and locations

General Document: ⭐⭐⭐ Good¶

Success rate: 50-70%

Why it's challenging: - Unstructured content - Implicit timelines - Mixed topics - Varied date formats

Typical extraction: - 2-10 events per document - 50-70% confidence scores - Many date ranges missing - Locations often absent

Optimization tips: - Add explicit dates where possible - Structure into sections - Use clear headers - Separate events with spacing

Document Preparation Best Practices¶

Date Formatting¶

Optimal formats:

✅ Excellent:
- January 2020 - March 2023
- Jan 2020 - Mar 2023
- 01/2020 - 03/2023
- 2020-01 to 2023-03

✅ Good:
- 2020 - 2023 (year only)
- Q1 2020 - Q4 2023
- Spring 2020 - Summer 2023

⚠️ Acceptable but less optimal:
- "Three years" (requires reference point)
- "Recently" (too vague)
- "2020s" (decade reference)

❌ Avoid:
- "a while ago"
- "during college"
- "around that time"

Section Headers¶

Recommended headers for resumes:

Work Experience / Professional Experience / Employment History
Education / Academic Background
Projects / Key Projects
Publications / Research
Skills / Technical Skills
Certifications / Licenses
Awards / Honors / Achievements

Headers ChronoScope recognizes: - Experience, Work, Employment, Career - Education, Academic, School, University - Projects, Portfolio, Work Samples - Publications, Research, Papers - Awards, Honors, Achievements, Recognition

Content Structure¶

Example: Well-structured resume entry

Senior Software Engineer                    ← Job title (clear)
Google LLC, Mountain View, CA              ← Company, Location
June 2020 - Present                        ← Dates (explicit)

• Led team of 5 engineers                  ← Bullet points
• Built ML pipeline handling 1M+ events    ← Quantified achievements
• Deployed to production in 6 months       ← Specific timeline

Example: Poorly-structured entry

Worked at Google for a while doing engineering stuff.
Was part of a team. Did some machine learning work.

File Size Limitations¶

Recommended Sizes¶

Handling Large Files¶

If file > 10 MB:

1. Compress PDF:

   # Using Ghostscript
   gs -sDEVICE=pdfwrite -dCompatibilityLevel=1.4 \
      -dPDFSETTINGS=/ebook -dNOPAUSE -dQUIET -dBATCH \
      -sOutputFile=output.pdf input.pdf
   

2. Split into sections:

3. Upload work experience separately
4. Upload education separately

5. Merge in ChronoScope

6. Remove images:

7. Use text-only export
8. Remove embedded images
9. Keep only textual content

Multi-Language Support¶

Supported Languages¶

Best Practices for Non-English¶

For best results:

1. Use ISO date format: YYYY-MM-DD
2. Translate key sections: Experience, Education
3. Keep location names in English: "Paris, France" vs "Paris, Francia"
4. Use English month names when possible: "Jan 2020" vs "Janv 2020"

Example (mixed language):

Ingénieur Logiciel Senior          ← French title (OK)
Google LLC, Paris, France           ← English location (Good)
January 2020 - Present              ← English dates (Best)

Troubleshooting by Document Type¶

PDFs Not Extracting¶

Low Confidence Scores¶

Missing Events¶

Duplicate Events¶

Document Type Comparison¶

Recommendations by Use Case¶

Career Portfolio¶

Best document types: - Professional resume (primary) - LinkedIn export (supplementary) - Cover letters (for narrative context)

Avoid: - General documents - Informal notes

Academic Timeline¶

Best document types: - Academic CV (primary) - Personal statements (secondary) - Publication lists (supplementary)

Avoid: - Non-academic documents - Mixed-purpose files

Life Story¶

Best document types: - Chronological life summary (create for this purpose) - Personal statements - Diary exports (if structured)

Avoid: - Unstructured journal entries - Stream-of-consciousness writing

Next Steps¶

After understanding document types:

• 📤 Upload Your Documents
• 📊 Visualize Your Timeline
• ✅ Validate Extracted Events
• 📝 View Your Timeline

Quick Reference¶

Back to Documentation Home